Integrated computational and preclinical evaluation of drug repurposing candidates for KRAS-mutant non-small cell lung cancer Authors: Anisha Jain, Chandan Dharmashekar, Bhargav Shreevatsa K S, Kavitha Raj, Amogha Shivakumar, Shiva Prasad Kollur, Chandan Shivamallu, Ashwini Prasad, M. N. Nagendra Prasad Published: 2026 ABSTRACT KRAS mutations, particularly KRAS G12C, pose significant therapeutic challenges in non-small cell lung cancer (NSCLC) due to limited inhibitor efficacy and rapid resistance. Drug repurposing offers a time and cost-effective strategy to identify alternative therapeutic options targeting KRAS-driven signalling. A structure-based drug repurposing approach was employed to screen FDA-approved compounds targeting key regulators to the MAPK signalling pathway and computationally modelled for potential engagement with the KRAS G12C switch-II pocket. Computational analyses including virtual screening, molecular docking, and molecular dynamics simulations were used to evaluate binding affinity, stability and potential protein-ligand interactions. Lead candidates were subsequently validated using in vitro functional assays, ex ovo chorioallantoic membrane (CAM) and in vivo chemical induced lung carcinoma model. Computational analyses identified nilotinib and risperidone as high affinity candidates across MAPK pathway targets. Functional assays confirmed that nilotinib markedly reduced NSCLC cell viability, colony formation and angiogenesis, while inhibiting cell migration and inducing apoptosis. In vivo, nilotinib significantly lowered tumor burden and modulated immune and inflammatory profiles in a chemically induced murine lung carcinoma. These findings support nilotinib as a promising therapeutic candidate for KRAS-mutant NSCLC. Further studies involving target-engagement assays, pathway-validation experiments and genetically defined KRAS-G12C driven models are warranted to establish its precise mechanism of action.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-60339-7. Introduction Lung cancer remains the leading cause of cancer-related mortality worldwide, accounting for 18.7% of all cancer deaths in 20221. It is the most frequently diagnosed cancer in men (15.2%) and the second in women (9.4%). In India, incidence and mortality rates were 5.32 and 5.79 per 100,000 population, respectively in 20221,2. Despite therapeutic advances, the overall 5-year survival rate remains below 20%, underscoring the need for novel treatment strategies. Non-small cell lung cancer (NSCLC) constitutes 80–85% of lung cancers, with adenocarcinoma, squamous cell carcinoma, and large cell carcinoma as major subtypes. Over 70% of patients are diagnosed at an advanced stage3,4. Smoking is the principal risk factor, although air pollution, radon, asbestos, diesel exhaust, and genetic predisposition also contribute to carcinogenesis5. The introduction of targeted therapies and immunotherapies has improved outcomes; however, resistance remains inevitable6. The Mitogen-Activated Protein Kinase (MAPK) signalling pathway comprising receptor tyrosine kinases (RTKs), RAS, RAF, MEK and ERK, is a central driver of proliferation, metastasis, and immune evasion (Supplementary Fig. 1)7. KRAS mutations occur in ~ 30% of adenocarcinomas, with KRAS G12C being the most common hotspot. EGFR alterations and BRAF mutations (notably V600E) also contribute to MAPK hyperactivation and MEK and ERK dysregulation further amplify tumour growth and confer resistance to upstream inhibitors8,9. In the GTP-bound ‘on’ state, KRAS activates downstream signalling pathways, mainly the MAPK/ERK pathway and the PI3K/AKT/mTOR pathway, leading to cell division and proliferation9,10. The approval of KRAS G12C inhibitors (sotorasib and adagrasib) has validated KRAS as a therapeutic target, however, acquired resistance through secondary KRAS mutations, RTK reactivation, and bypass signalling limits their clinical benefit9,10. These inhibitors engage the mutant Cys12 residue covalently within the switch-II pocket, locking KRAS in its GDP-bound inactive state and disrupting mediated nucleotide exchange. Co-targeting upstream molecules such as SHP2, which mediates RTK–KRAS signalling, has emerged as a rational strategy to enhance the durability of KRAS blockade11. Several SHP2 and MEK inhibitors are currently in clinical development, reflecting the translational potential of multi-node MAPK pathway inhibition10–12. Conventional drug discovery is expensive and time-intensive, with only ~ 10% of candidates achieving regulatory approval4,13. Drug repurposing offers a pragmatic alternative by leveraging existing pharmacological and safety profiles to accelerate development while reducing costs14. Nearly 30% of FDA-approved drugs and vaccines in recent years were found using in silico methods14,15. Structure-based computational approaches enable identification of approved or investigational drugs with novel activity against oncogenic drivers16. In this study, we employed an in silico drug repurposing pipeline to computationally evaluate candidate inhibitors targeting KRAS G12C and key regulators within the MAPK pathway (Fig. 1). Top-ranked compounds were further evaluated for anticancer activity in KRAS G12C-mutant NSCLC cell lines and validated ex ovo and in vivo, providing a translational framework for accelerating therapeutic discovery against this challenging target. Fig. 1An overview of the structure-based drug repurposing pipeline employed to identify lead candidates for NSCLC from an FDA approved drug library. The figure was created in Microsoft PowerPoint 365. An overview of the structure-based drug repurposing pipeline employed to identify lead candidates for NSCLC from an FDA approved drug library. The figure was created in Microsoft PowerPoint 365. Methodology Preparation of NSCLC-associated mutant target proteins and FDA approved ligand library Computational structure-based drug discovery was conducted using the Schrödinger Suite 2023-1 (Maestro-v13.3, Glide-v9.6, Desmond-v7.1)15,16 accessed via server-based license installed at JSS AHER, Mysuru. The three-dimensional structures of the target proteins were retrieved from the RCSB Protein Data Bank (PDB)17. Table 1 lists the proteins based on the literature review and structural parameters such as experimental resolution, mutation information and co-crystalized ligands. These structures were prepared using the Protein Preparation Wizard of the suite by assigning bond orders, adding missing hydrogens, adding disulfide bonds, filling in missing loops, removing water molecules, ions/metals and co-crystalized ligands16. FDA-approved drug library obtained from Selleckchem (Catalogue No.L1300) comprises of a unique collection of 3183 approved drugs for HTVS. LigPrep preprocessing was performed using the OPLS4 forcefield with generation of relevant ionization and tautomeric states at physiological pH (7.0 ± 2.0) generating 3984 ligand structures for HTVS against the target structures18–20. Table 1Structural information of target proteins of the MAPK pathway acquired from Protein Data Bank.Sl. no.Target protein namePDB IDExperimental informationCo-existing ligandResolution (Å)Ramachandran plot favoured residues (%)1.EGFR tyrosine kinase domain4HJO212.7599%Erlotinib2.Human KRAS G12C6OIM221.6598%Sotorasib3.B-Raf Kinase V600E3OG7232.4594%Vemurafenib4.Human MEK14AN2242.5092%Cobimetinib5.Human ERK14QTB251.4098%SCH7729846.Non-receptor protein tyrosine phosphatase SHP25EHR261.7097%SHP099 Table 1Structural information of target proteins of the MAPK pathway acquired from Protein Data Bank.Sl. no.Target protein namePDB IDExperimental informationCo-existing ligandResolution (Å)Ramachandran plot favoured residues (%)1.EGFR tyrosine kinase domain4HJO212.7599%Erlotinib2.Human KRAS G12C6OIM221.6598%Sotorasib3.B-Raf Kinase V600E3OG7232.4594%Vemurafenib4.Human MEK14AN2242.5092%Cobimetinib5.Human ERK14QTB251.4098%SCH7729846.Non-receptor protein tyrosine phosphatase SHP25EHR261.7097%SHP099 Structural information of target proteins of the MAPK pathway acquired from Protein Data Bank. Structure-based high throughput virtual screening (HTVS) Virtual screening involves the use of computational tools to dock potential drug candidates against protein structures and identify compounds with high likelihood of binding. A grid box was defined around the co-crystalized ligand binding site using the Receptor Grid Generation tool with a scaling factor of 1 and partial charge cutoff of 0.2516. Ligand-flexible docking was performed using the HTVS precision mode followed by SP (standard precision) mode using the Glide module. After thorough filtering based on docking score, non-bonded interactions, geometric ligand-binding site complementarity literature evidence of anticancer activity and suitability for repurposing, top three hits were selected for SP docking26,27. To computationally assess the potential for covalent engagement with the reactive Cys12 residue of KRAS G12C, top hits were evaluated using the CovDock panel in Glide. Nucleophilic addition to a double bond was selected as the reaction type to model potential Michael acceptor-type chemistry, consistent with the mode of action of approved KRAS G12C inhibitors such as sotorasib. The covalent docking scores and interaction geometries were used as a comparative metrics alongside non-covalent docking results. Prime VSGB2.0 energy model was employed to minimize and score the top-ranked covalent complexes28. Molecular dynamics (MD) simulation MD simulations were conducted using the Desmond module of the Schrödinger Suite to assess the stability and intermolecular interactions of protein-ligand complexes over time26. Using the System Builder tool, each docked complex was placed in an orthorhombic simulation box with a buffer distance of 14 Å, solvated with the SPC water model (Supplementary Fig. 2), neutralized with Na+/Cl− ions and equilibrated at pH 7.0 ± 0.2 under NPT ensemble conditions (1.01325 bar, 310 K)29. Initial MD simulations for all six MPK pathway targets docked with nilotinib (C1) were run for 100 ns. For the KRAS G12C-nilotinib covalent complex, an extended 300 ns simulation was performed using the TIP3P water model to capture longer-timescale conformational dynamics and to obtain a more detailed interaction profile. The use of TIP3P is consistent with widely adopted practices for covalent complex simulations29. Simulation trajectories were analysed using the Simulation Interaction Diagram panel and the graphs were assembles using QtGrace v0.2.616. Based on the above analysis, two drug compounds were taken for further in vitro analysis. Maintenance of cell lines and compounds NCI-H23 (KRAS G12C-mutant) cells were kindly provided by Adichunchanagiri Institute of Medical Sciences, Karnataka and were cultured in RPMI-1640 medium supplemented with 10% FBS, 1% penicillin–streptomycin and 1% amphotericin B at 37 °C in 5% CO2. Nilotinib (#CDS023093) and risperidone (#S1615) were purchased from Selleck Chemicals, dissolved in DMSO, aliquoted and stored at -80 °C until use. All reagents were of cell biology grade and purchased from Sisco Research Laboratories (SRL) India, Sigma Aldrich, USA and Thermo Scientific. Assessment of cell cytotoxicity and proliferation using MTT colorimetric assay Cell viability was assessed using the MTT colorimetric assay. NCI H23 cells were seeded at 2 × 104 cells/well in Corning 96-well plates (Sigma) and treated with compound concentrations ranging from 2 to 200 µM for 24 h with cisplatin (50 µM) as the positive control. MTT reagent was added and metabolized to formazan by viable cells. The resulting crystals were dissolved in DMSO, and absorbance measured at 550 nm using a microplate reader (Infinite M Plus, Tecan, Switzerland)30. Clonogenic assay The ability of single cells to form colonies following drug exposure was assessed. Cells were treated with IC50 concentrations of the compounds for 72 h and subsequently allowed to recover in fresh medium for 10 days. Colonies were fixed with paraformaldehyde, stained with crystal violet, observed under light microscope at 10x and 20x magnifications and the colony numbers were quantified using ImageJ software31. This assay assessed the sustained antiproliferative potential of the compounds. The drug compound showing the best results from the above two assays was considered for further evaluation. Scratch wound healing assay Cell migration was evaluated using the scratch wound healing assay. Cells were seeded in 6 well tissue culture plates to confluency and a straight wound were introduced along the cell monolayer using a sterile 200 µl pipette tip. Cells were treated with the IC50 concentration of the nilotinib, and wound closure was monitored at 0, 24 and 48 h. Wound area was quantified using ImageJ software31,32. In this assay, the percentage wound confluence represents the proportion of the initial wound area that was closed by cell migration, calculated as: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{Wound}}\;{\mathrm{confluence}}\;(\% ) = \left( {{\mathrm{A}} - {\mathrm{B}}} \right){\mathrm{/A}} \times 100$$\end{document} Where, A is Initial wound area and B is Final wound area (24–48 h). A lower wound confluence percentage indicated greater inhibition of cell migration while a higher percentage indicated more extensive migration-driven wound closure. FACS cell cycle analysis Cell cycle distribution was evaluated by flow cytometry with propidium iodide (PI) staining. Cells were seeded at 2 × 105 cells/well in 6-well plates and treated at IC50 concentration for 24 h. Following fixation and PI staining, DNA content was analyzed to determine cell distribution across G0/G1, S, and G2/M phases32,33. Accumulation in a specific phase was interpreted as indicative of cell cycle arrest at the checkpoint. Annexin V-FITC apoptosis assay Apoptosis was quantified using the FITC-Annexin V apoptosis detection kit. Cells were seeded in 6-well plates at 1 × 105 cells/well and incubated for 24 h before treatment. Following treatment, cells were harvested, washed, resuspended in 500 µl of Annexin V binding buffer and stained with 5 µl of FITC Annexin V and 5 µl PI and analysed by flow cytometry32,33. Evaluation of anti-angiogenic activity using the ex vivo CAM assay Pathogen-free embryonated chicken eggs (3-day old) were obtained from a certified hatchery in Mysuru, Karnataka, India. The ex ovo CAM assay was performed according to the protocol described by Naik et al.34. On embryonic development day (EDD) 4, eggs were cracked open under sterile conditions and contents transferred to sterile weigh boats and covered with sterile cling wrap. On day 9, sterile filter discs soaked in conditioned medium from treated or untreated cells (prepared as per established previous protocol) were placed over chorioallantoic membrane (CAM) and incubated for further 5 days34. Embryos that failed to survive until the study endpoint or exhibited developmental abnormalities unrelated to treatment were excluded from analysis. On EDD 14, CAM images were captured and analyzed using AngioTool software to quantify average total vessel length and number of branching junctions. Image analysis was performed using identical analysis parameters for all groups. On day 15, embryos were sacrificed and tumor masses were isolated, fixed in 4% formalin and subjected to histopathological analysis35. This model provided an ex vivo-like platform for evaluating anti-angiogenic potential. In vivo analysis in the BALB/c benzo[a]pyrene-induced lung carcinogenesis model BALB/c mice (20–25 g) were procured from KCC Bio-labs, Tumkur, Karnataka and maintained under standard conditions in compliance with Committee for Control and Supervision of Experiments on Animals (CPCSEA) guidelines, with prior approval from the Institutional Animal Ethics Committee (IAEC) of KCC Bio-labs, Tumur, Karnataka and all procedures for anaesthesia/euthanasia and the experiments were performed in accordance with CPCSEA guidelines and regulations. For procedural interventions, animals were anesthetised using a combined protocol of ketamine (60 mg/kg) and xylazine (7 mg/kg). Lung carcinogenesis was induced using benzo[a]pyrene (B[a]P, 50 mg/kg body weight dissolved in 0.5% DMSO and corn oil). Nilotinib was administered orally at 50 and 75 mg/kg B.W. Four groups of six mice each were studied: Group I (corn oil, vehicle control), Group II (B[a]P only), Group III (B[a]P + nilotinib 50 mg/kg), and Group IV (B[a]P + nilotinib 75 mg/kg). Following the experimental period, blood and lung samples were collected for biochemical assessment of tumour-associated markers (ADA, AHH, LDH) and histopathological analysis. At study completion, animals were euthanised by CO2 inhalation followed by cervical dislocation to confirm death, in accordance with CPCSEA. Lung tissues were fixed in formalin, paraffin-embedded, sectioned, and stained with haematoxylin and eosin (H&E). Relative lung weight (RLW) was calculated as an indirect surrogate of tumour burden36,37. Flow cytometry analysis of inflammatory markers Single-cell suspensions from lung tissues were prepared and immunophenotyped using fluorescently labelled monoclonal antibodies: BV510 anti-Ly6C, PerCP-Cy5 anti-CD3, AF647 anti-CD11c, FITC anti-CD45, MR6F3 anti-CD206, BM8 anti-F4/80 and BV786 anti-CD4. Samples were analyzed using a BD FACS Celesta flow cytometer and FCS (Flow Cytometry Standard) files were processed with FlowJo software (BD Biosciences, New Jersey, USA)38. Results HTVS analysis An FDA approved drug library taken from Selleckchem was screened against the co-crystalized ligand binding sites of the MAPK pathway target proteins using the Glide HTVS module. A graphical representation of the Glide Gscore (kcal/mol) obtained from HTVS is presented in Fig. 2. Because Glide docking scores correlates with the free energy of binding, more negative value indicates stronger predicted binding affinity. Compounds were filtered based on negative scores (threshold >- 7 kcal//mol), receptor-ligand binding complementarity, standard inhibitors, and we also filtered the compounds that are the subject of ongoing study. The top three hits were reevaluated by SP docking. Fig. 2A graphical representation of the glide Gscores from HTVS of the FDA approved drug library against MAPK pathway target proteins- (A) EGFR and KRAS, (B) BRAF and MEK, (C) ERK and SHP2. A graphical representation of the glide Gscores from HTVS of the FDA approved drug library against MAPK pathway target proteins- (A) EGFR and KRAS, (B) BRAF and MEK, (C) ERK and SHP2. Standard precision molecular docking Table 2; Fig. 3 summarize the SP molecular docking of the top three hits across all sic. target proteins relative to their respective standard inhibitors. Supplementary Fig. 3 represents the geometric ligand-binding site complementarity of the compounds docked to the binding pockets highlighted in grey color. Table 2Standard precision molecular docking details of the compounds C1-C3 relative to their respective standard inhibitors docked to the target proteins of the MAPK pathway.Sl. no.Target proteinCompound IDGlide score (kcal/mol)Hydrogen bondsTotal interactionsActive site residues forming interactions1EGFRC1− 8.421 salt bridge, 1 pi-pi stackingCYS 773, ASP 776, GLU 780C2− 7.811 salt bridgePHE 699, ASP 613C3− 7.3521 salt bridge, 2 pi-cationLYS 721, ASN 818Standard− 9.112–MET 769, CYS 7732KRASC1− 9.8421 salt bridge, 1 pi-cation,1 pi-pi StackingCYS 12, PRO 34, ARG 68, TYR 96C2− 8.212 pi-cationGLN 61, ARG 68, GLN 99C3− 7.511 salt bridgeCYS 12, ALA 59Standard− 9.221 salt bridgeCYS 12, GLN 61, GLU 633BRAFC1− 7.821 salt bridgeCYS 532, ASP 594C2− 8.83–CYS 532, SER 536C3− 7.1731 pi-cationLYS 481, TRP 511, CYS 532Standard− 7.482–GLN 530, CYS 5324MEKC1− 8.5721 salt bridgeARG 189C2− 9.253–ASN 79, LYS 192, ASP 208C3− 8.521 pi-pi stackingASP 208, PHE 209Standard− 9.2811 pi-pi stackingARG 189, PHE 2295ERKC1− 8.4722 salt bridgeLYS 71, ASP 128 LYS 168C2− 7.831 salt bridgeGLU 83, CYS 183, ASP 184C3− 6.811 salt bridgeLYS 71, ASN 171, ASP 184Standard− 6.9431 halogen bond, 1 pi-pi StackingTYR 53, LYS 71, ASP 1846SHP2C1− 8.8912 salt bridge, 1 pi-pi StackingARG 111, GLU 232, GLU 250C2− 8.312 salt bridge, 1 pi-cationGLU 249, GLU 250, PHE 114C3− 7.212–ARG 111, THR 218Standard− 7.522 salt bridgeGLU 249 Table 2Standard precision molecular docking details of the compounds C1-C3 relative to their respective standard inhibitors docked to the target proteins of the MAPK pathway.Sl. no.Target proteinCompound IDGlide score (kcal/mol)Hydrogen bondsTotal interactionsActive site residues forming interactions1EGFRC1− 8.421 salt bridge, 1 pi-pi stackingCYS 773, ASP 776, GLU 780C2− 7.811 salt bridgePHE 699, ASP 613C3− 7.3521 salt bridge, 2 pi-cationLYS 721, ASN 818Standard− 9.112–MET 769, CYS 7732KRASC1− 9.8421 salt bridge, 1 pi-cation,1 pi-pi StackingCYS 12, PRO 34, ARG 68, TYR 96C2− 8.212 pi-cationGLN 61, ARG 68, GLN 99C3− 7.511 salt bridgeCYS 12, ALA 59Standard− 9.221 salt bridgeCYS 12, GLN 61, GLU 633BRAFC1− 7.821 salt bridgeCYS 532, ASP 594C2− 8.83–CYS 532, SER 536C3− 7.1731 pi-cationLYS 481, TRP 511, CYS 532Standard− 7.482–GLN 530, CYS 5324MEKC1− 8.5721 salt bridgeARG 189C2− 9.253–ASN 79, LYS 192, ASP 208C3− 8.521 pi-pi stackingASP 208, PHE 209Standard− 9.2811 pi-pi stackingARG 189, PHE 2295ERKC1− 8.4722 salt bridgeLYS 71, ASP 128 LYS 168C2− 7.831 salt bridgeGLU 83, CYS 183, ASP 184C3− 6.811 salt bridgeLYS 71, ASN 171, ASP 184Standard− 6.9431 halogen bond, 1 pi-pi StackingTYR 53, LYS 71, ASP 1846SHP2C1− 8.8912 salt bridge, 1 pi-pi StackingARG 111, GLU 232, GLU 250C2− 8.312 salt bridge, 1 pi-cationGLU 249, GLU 250, PHE 114C3− 7.212–ARG 111, THR 218Standard− 7.522 salt bridgeGLU 249 Standard precision molecular docking details of the compounds C1-C3 relative to their respective standard inhibitors docked to the target proteins of the MAPK pathway. 1 salt bridge, 1 pi-cation, 1 pi-pi Stacking Fig. 3Glide Gscores from SP docking of the top three hits bound to the MAPK pathway target proteins. Standard inhibitors of each target are shown for comparison. Glide Gscores from SP docking of the top three hits bound to the MAPK pathway target proteins. Standard inhibitors of each target are shown for comparison. The data highlights that C1 consistently exhibited strong binding affinities across most targets, surpassing or equalling the standard inhibitors, particularly against KRAS, ERK and SHP2. C2 also demonstrated favorable docking against BRAF and MEK. C3 showed weaker binding compared with C1 and C2 but retained key interactions like salt bridges and pi-cation bonds. Protein-ligand binding is widely thought to be facilitated by H-bonds. One of the strongest bonds is a salt bridge, which is a non-covalent interaction between two ionized sites. Overall, C1 emerged as the most promising candidate, based on its combination of strong glide scores and diverse stabilizing interactions, across the MAPK target panel. C1 docked to all the target proteins were further considered for MD simulations. CovDock analysis CovDock analysis was performed on the top three hits using sotorasib as a reference covalent inhibitor to assess the potential for engagement with the KRAS G12C Cys12 residue. All three compounds modelled a covalent interaction with Cys12 under the reaction type ‘nucleophilic addition to a double bond’ (Table 3; Fig. 4). The KRAS G12C cryptic pocket residues His95/Tyr96/Gln99 shown to be important for clinical inhibitor development were found to interact with C1, with a CovDock score of -9.01 kcal/mol, comparable to sotorasib (-9.4 kcal/mol). The CovDock analysis models the reaction geometry and scores the resulting complex but does not confirm that bond formation occurs in biological systems. The information of the top three lead compounds screened from the computational assays are presented in Table 4. Table 3Computationally covalent docking details for top hits against KRAS G12C.Compound IDCovDock score (kcal/mol)Covalent bond distance (Å)H-bond forming residuesTotal interactionsC1− 9.012.25GLN 994C2− 7.172.25CYS 12, LYS 88, GLN 994C3− 6.51.96CYS 12, TYR 963Sotorasib− 9.42.11CYS 12, GLN 61, GLN 994 Table 3Computationally covalent docking details for top hits against KRAS G12C.Compound IDCovDock score (kcal/mol)Covalent bond distance (Å)H-bond forming residuesTotal interactionsC1− 9.012.25GLN 994C2− 7.172.25CYS 12, LYS 88, GLN 994C3− 6.51.96CYS 12, TYR 963Sotorasib− 9.42.11CYS 12, GLN 61, GLN 994 Computationally covalent docking details for top hits against KRAS G12C. Fig. 4Two-dimensional protein-ligand interactions maps illustrating computationally predicted binding modes of C1-C3 docked to KRAS G12C via CovDock in comparison with Sotorasib. Two-dimensional protein-ligand interactions maps illustrating computationally predicted binding modes of C1-C3 docked to KRAS G12C via CovDock in comparison with Sotorasib. Table 4Information of the top lead compounds screened for further assays.Sl. no.Compound IDCompound namePubChem IDDrug classOriginal indication1.C1Nilotinib644,241Tyrosine kinase inhibitorTreatment of chronic myelogenous leukaemia (Philadelphia chromosome-positive)392.C2Risperidone5073Antipsychotic (Pyridopyrimidine)Schizophrenia and bipolar disorder403.C3Loperamide3955Anti-diarrheal agent (mu-opioid receptor agonist)Diarrhoea management41 Table 4Information of the top lead compounds screened for further assays.Sl. no.Compound IDCompound namePubChem IDDrug classOriginal indication1.C1Nilotinib644,241Tyrosine kinase inhibitorTreatment of chronic myelogenous leukaemia (Philadelphia chromosome-positive)392.C2Risperidone5073Antipsychotic (Pyridopyrimidine)Schizophrenia and bipolar disorder403.C3Loperamide3955Anti-diarrheal agent (mu-opioid receptor agonist)Diarrhoea management41 Information of the top lead compounds screened for further assays. MD simulation analysis Root mean square deviation (RMSD) trajectories over 100 ns were analysed for C1 docked to all six MAPK pathway targets. RMSD measures the deviation in protein backbone atom positions over time relative to the starting structure. Lower and stable RMSD values indicate conformational stability of the complex. As shown in Fig. 5, RMSD values for the EGFR, BRAF and SHP2 complexes fluctuated initially (0–10 ns) before stabilizing throughput the remainder of the trajectory. The MEK complex fluctuated at 38 ns with RMSD ranging from 3 Å to 5 Å though the ligand RMSD remained comparatively stable. The ERK Complex, the protein showed minor fluctuations at 25 ns and 80 ns with ligand RMSD stabilizing from 60 to 100ns approximately 2 Å. For KRAS G12C complexes (Fig. 6), the protein RMSD fluctuated throughout the 100ns between 1 and 3 Å. C1 initially fluctuated to 2 Å within the first 5 ns and then stabilized for the rest of the simulation time. This complex showed better stability when compared to the RMSDs of the other compounds. C2 deviated at 38 ns and remained relatively stable. C3 displayed continuous deviations throughout the run time making the RMSD less satisfactory for the KRAS G12C. The extended 300 ns simulation time of the KRAS G12C-C1 covalent complex (Fig. 7) showed initial fluctuations of 1.5–2.5 Å within the first 50 ns, with stable trajectories thereafter. Hydrogen bonding with Cys 12 was maintained for approximately 100% of the simulation time and multiple specific contacts were observed with residues including Gln99, Val9, Lys16 and Thr58 (Fig. 7B-C). RMSF analysis (Fig. 7D) revealed the flexible loop regions accounted for the majority of backbone fluctuation, while the binding pocket residues remained stable throughput. Fig. 5RMSD analysis of C1 (orange) complexed with KRAS G12C, EGFR, BRAF, MEK, ERK and SHP2 over 100 ns with respect to RMSD of apo-protein (purple). RMSD analysis of C1 (orange) complexed with KRAS G12C, EGFR, BRAF, MEK, ERK and SHP2 over 100 ns with respect to RMSD of apo-protein (purple). Fig. 6RMSD trajectories determining the fluctuations and stability of KRAS unbound protein and complexed with the compounds C1 to C3 in comparison with sotorasib. The RMSD graphs are colour coded differentiating the compounds on a 100ns time frame. RMSD trajectories determining the fluctuations and stability of KRAS unbound protein and complexed with the compounds C1 to C3 in comparison with sotorasib. The RMSD graphs are colour coded differentiating the compounds on a 100ns time frame. Fig. 7Extended MD Simulation analysis of 300 ns for C1- KRAS G12C computationally predicted covalent complex considering the parameters (A) Protein backbone-Ligand fit RMSD, (B) Stacked-bar chart of protein-ligand interaction fractions, and (C) timeline of per-residue contact frequency with darker orange indicating higher contact frequency, (D) Protein RMSF (red shading: α-helix regions; blue shading: β-strand regions; green vertical bars: residues with ligand contacts). Extended MD Simulation analysis of 300 ns for C1- KRAS G12C computationally predicted covalent complex considering the parameters (A) Protein backbone-Ligand fit RMSD, (B) Stacked-bar chart of protein-ligand interaction fractions, and (C) timeline of per-residue contact frequency with darker orange indicating higher contact frequency, (D) Protein RMSF (red shading: α-helix regions; blue shading: β-strand regions; green vertical bars: residues with ligand contacts). These simulation results support the computational stability of CI-KRAS G12C complex and the broader MAPK target complexes. Based on the comprehensive in silico analysis, nilotinib (C1) and risperidone (C2) were carried forward for in vitro experimental validation. Assessment of cell viability using MTT colorimetric assay The cytotoxic activity of nilotinib (C1) and risperidone (C2) was assessed in KRAS G12C-mutant NCI-H23 cells using the MTT assay (Fig. 8). Both compounds demonstrated clear dose-dependent reductions in cell viability. Nilotinib showed markedly greater cytotoxic potency than risperidone, as reflected by its substantially lower IC50 value (31.51 µM vs. 81.78 µM). Although the IC50 of nilotinib is higher than that of approved KRAS G12C inhibitors such as sotorasib (sub-micromolar range), this difference is not unexpected given that nilotinib was not purpose-designed for KRAS G12C targeting. Accordingly, risperidone is not carried forward beyond the clonogenic survival assay, and nilotinib was confirmed as the lead compound for subsequent characterisation. Fig. 8NCI-H23 cells were treated with the lead compounds (2-200 µM) for 24 h, with cisplatin as the standard control. Cell viability is expressed as a percentage of untreated control. Data are shown as mean ± SEM (n = 3); statistical analysis was performed using one-way ANOVA (p < 0.05). NCI-H23 cells were treated with the lead compounds (2-200 µM) for 24 h, with cisplatin as the standard control. Cell viability is expressed as a percentage of untreated control. Data are shown as mean ± SEM (n = 3); statistical analysis was performed using one-way ANOVA (p < 0.05). Clonogenic cell survival analysis The antiproliferative efficacy of the lead compounds was assessed by colony formation assay (Fig. 9). Nilotinib significantly reduced the average colony count to 233 ± 7, compared with 396 ± 6 for risperidone, 351 ± 10 for cisplatin (standard control), and 557 ± 12 for untreated cells (p < 0.05, one-way ANOVA). These findings reinforce the superior long-term antiproliferative efficacy of nilotinib and validate its selection as the lead compound for further experimental studies. Hence, nilotinib was considered as the lead compound. Fig. 9(A) Six-well plate images of the stained colonies from the untreated cells, standard cisplatin, along with two lead compounds. (B) Quantification of average colony count per group. Experiment was performed in triplicates. Data are shown as mean ± SEM (n = 3); significance was determined using one-way ANOVA (p < 0.05). **** p < 0.0001 vs. untreated (C). Representative colony morphology images 10x and 20x magnifications. (A) Six-well plate images of the stained colonies from the untreated cells, standard cisplatin, along with two lead compounds. (B) Quantification of average colony count per group. Experiment was performed in triplicates. Data are shown as mean ± SEM (n = 3); significance was determined using one-way ANOVA (p < 0.05). **** p < 0.0001 vs. untreated (C). Representative colony morphology images 10x and 20x magnifications. Cell migration by wound healing assay The effect of nilotinib on collective cell migration was evaluated using the scratch wound healing assay in NCI-H23 cells (Fig. 10). A lower wound closure percentage indicates greater inhibition of cell migration. At 24 h, nilotinib-treated cells exhibited 30.82% wound closure, compared with 40.86% for cisplatin-treated cells. At 48 h, wound closure was 45.07% for nilotinib versus 47.92% for cisplatin. Values for both nilotinib and cisplatin at 48 h converge, suggesting that longer-term migration inhibition is comparable between the two agents Notably, nilotinib exhibited a slightly more pronounced effect, indicating impaired collective cell migration, a key process underlying tumor invasion and metastasis. Fig. 10Quantitative scatter plot depicts percentage would closure. Data are presented as mean ± SEM from three independent experiments performed in triplicate (n = 3). Statistical significance was determined using one-way ANOVA (p < 0.05). Quantitative scatter plot depicts percentage would closure. Data are presented as mean ± SEM from three independent experiments performed in triplicate (n = 3). Statistical significance was determined using one-way ANOVA (p < 0.05). Cell cycle study using FACS The effect of nilotinib on cell cycle distribution in NCI-H23 cells was assessed by PI staining and flow cytometry (Fig. 11). In untreated cells, the distribution was: sub-G0/G1 2.3%, G0/G1 61.39%, S phase 14.21%, and G2/M 22.1%. Cisplatin treatment shifted the distribution to sub-G0/G1 6.96%, G0/G1 52.14%, S phase 10.06%, and G2/M 30.84%. Following treatment with nilotinib at IC₅₀, the distribution was: sub-G0/G1 2.19%, G0/G1 50.85%, S phase 13.27%, and G2/M 33.69%. Relative to untreated cells, nilotinib treatment produced a decrease in the G0/G1 population with a concomitant increase in the G2/M fraction and a modest decrease in the S-phase fraction. This pattern is consistent with G2/M-phase cell cycle accumulation, suggesting that nilotinib arrests cells at the G2/M checkpoint rather than at the G1/S transition. This is mechanistically plausible given that MAPK pathway dysregulation can interfere with CDC25C phosphatase activity and CDK1/cyclin B1-mediated G2/M progression. Fig. 11Flow cytometric histograms (x-axis: DNA content; y-axis: cell count) and quantitative analysis showing the distribution of NCI-H23 cells across sub-G0/G1, G0/G1, S, and G2/M phases after 24 h treatment with C1 compared with untreated and cisplatin-treated controls. Data are expressed as mean ± SEM (n = 3). Statistical significance was determined using one-way ANOVA with post-hoc testing (p < 0.05). Flow cytometric histograms (x-axis: DNA content; y-axis: cell count) and quantitative analysis showing the distribution of NCI-H23 cells across sub-G0/G1, G0/G1, S, and G2/M phases after 24 h treatment with C1 compared with untreated and cisplatin-treated controls. Data are expressed as mean ± SEM (n = 3). Statistical significance was determined using one-way ANOVA with post-hoc testing (p < 0.05). Annexin V-FITC apoptosis assay Apoptotic cell populations following nilotinib treatment were quantified by Annexin V-FITC/PI flow cytometry (Figs. 12 and 13). Apoptosis plots were divided into four quadrants corresponding: The upper left (UL) (Q1) showed necrotic cells (Annexin V(-), PI(+)), upper right (UR) (Q2) showed late apoptotic cells (Annexin V(+), PI(+)), lower left (LL) (Q3) showed viable cells (Annexin V(-), PI(-)), lower right (LR) (Q4) showed early apoptotic cells (Annexin V(+), PI(-)). Nilotinib significantly increased early apoptotic (51.63%) and late apoptotic (21.8%) cell fractions compared with untreated controls for a combined apoptotic fraction of approximately 73.4%. Cisplatin treated cells showed 63.99% apoptosis and 12.83% at 24 h. A higher proportion of necrotic cells was observed in nilotinib vs. cisplatin-treated cells. These findings indicate that nilotinib exerts a strong pro-apoptotic effect in NSCLC cells, likely reflecting disruption of MAPK-mediated survival signalling. Fig. 12Flow cytometric dot plots showing Annexin V-FITC (x-axis) vs. propidium iodide (y-axis) staining in NCI-H23 cells treated with C1, cisplatin or left untreated. Quadrant assignments: Q1 (upper left) = necrotic cells (Annexin V-/PI+); Q2 (upper right) = late apoptotic (Annexin V+/PI+); Q3 (lower left) = viable (Annexin V-/PI-); Q4 (lower right) = early apoptotic (Annexin V+/PI-). Flow cytometric dot plots showing Annexin V-FITC (x-axis) vs. propidium iodide (y-axis) staining in NCI-H23 cells treated with C1, cisplatin or left untreated. Quadrant assignments: Q1 (upper left) = necrotic cells (Annexin V-/PI+); Q2 (upper right) = late apoptotic (Annexin V+/PI+); Q3 (lower left) = viable (Annexin V-/PI-); Q4 (lower right) = early apoptotic (Annexin V+/PI-). Fig. 13(A) Quantitative bar graph of the percentage of cells in each apoptotic category (viable, early apoptotic, late apoptotic and necrotic) across the three groups (B) A cumulative representation of early and late apoptotic fraction (%) per treatment group. Data are presented as mean ± SEM from three independent experiments in triplicate (n = 3). Statistical significance was determined by one-way ANOVA. ####p < 0.0001 vs. untreated control (viable cells), ****p < 0.0001 vs. untreated control (late apoptotic cells). (A) Quantitative bar graph of the percentage of cells in each apoptotic category (viable, early apoptotic, late apoptotic and necrotic) across the three groups (B) A cumulative representation of early and late apoptotic fraction (%) per treatment group. Data are presented as mean ± SEM from three independent experiments in triplicate (n = 3). Statistical significance was determined by one-way ANOVA. ####p < 0.0001 vs. untreated control (viable cells), ****p < 0.0001 vs. untreated control (late apoptotic cells). Evaluation of anti-angiogenic activity using the CAM assay The ex ovo CAM assay was employed to assess anti-angiogenic activity following nilotinib treatment (Fig. 14). Induction with NSCLC conditioned medium increased capillary branch density and vessel thickness. AngioTool-based quantification demonstrated that nilotinib treatment significantly reduced the total number of vascular junctions and average vessel length relative to untreated controls. Although vessel percentage area was reduced compared with untreated controls, it was comparable to cisplatin, indicating that nilotinib primarily disrupts vascular complexity and branching rather than total vascular coverage. Fig. 14(A) Representation of CAM images at EDD 14, showing vascular response to untreated and treated conditioned medium discs. (B) Enlarged views of vascular architecture and branching patterns (arrows indicating branching sites). (C) AngioTool-processed images illustrating vessel skeletonization and branching point analysis (D) Quantitative comparison of vessel percentage area, total number of junctions and average vessel length across treatment groups. Data are presented as mean ± SEM (n = 3). Statistical significance was determined by one-way ANOVA. **p < 0.01, ****p < 0.0001 vs. untreated control; ns, not significant. (A) Representation of CAM images at EDD 14, showing vascular response to untreated and treated conditioned medium discs. (B) Enlarged views of vascular architecture and branching patterns (arrows indicating branching sites). (C) AngioTool-processed images illustrating vessel skeletonization and branching point analysis (D) Quantitative comparison of vessel percentage area, total number of junctions and average vessel length across treatment groups. Data are presented as mean ± SEM (n = 3). Statistical significance was determined by one-way ANOVA. **p < 0.01, ****p < 0.0001 vs. untreated control; ns, not significant. In vivo BALB/c analysis The effect of nilotinib on lung tumour development was assessed in a B[a]P-induced murine lung carcinogenesis model (Figs. 15 and 16; Table 5). The B[a]P model generates chemically induced, polyclonal lung tumours via oxidative DNA damage and inflammatory mechanisms. The in vivo results presented here therefore reflect the broader anti-tumour and anti-inflammatory activity of nilotinib. Relative lung weight was significantly elevated in Group II (B[a]P only) compared with controls and was reduced in a dose-dependent manner in nilotinib-treated groups (Groups III and IV), with the 75 mg/kg dose producing the greater reduction (Fig. 15). Fig. 15The percentage of relative lung weight in BALB/c mice across all four treatment groups. Statistical significance was determined by one-way ANOVA; ***p < 0.001, ****p < 0.0001 vs. group II. The percentage of relative lung weight in BALB/c mice across all four treatment groups. Statistical significance was determined by one-way ANOVA; ***p < 0.001, ****p < 0.0001 vs. group II. Fig. 16H&E-stained lung tissue sections (50x magnification) from all four treatment groups: (A) Group I (naïve control): normal lung architecture showing bronchioles (b) and alveolar sacs with thin septa (arrow). (B) Group II (B[a]P): well-defined tumor mass formed of irregular papillae lined by hyperplastic epithelial cells with peribronchiolar inflammatory infiltration. (C) Group III (B[a]P + Nilotinib 50 mg/kg B.W): Thickened septa (arrow) with residual inflammatory cellular infiltration. (D) Group IV (B[a]P + Nilotinib 75 mg/kg B.W): thinned septa (arrow) indicating substantially reduced inflammation and tumour burden. H&E-stained lung tissue sections (50x magnification) from all four treatment groups: (A) Group I (naïve control): normal lung architecture showing bronchioles (b) and alveolar sacs with thin septa (arrow). (B) Group II (B[a]P): well-defined tumor mass formed of irregular papillae lined by hyperplastic epithelial cells with peribronchiolar inflammatory infiltration. (C) Group III (B[a]P + Nilotinib 50 mg/kg B.W): Thickened septa (arrow) with residual inflammatory cellular infiltration. (D) Group IV (B[a]P + Nilotinib 75 mg/kg B.W): thinned septa (arrow) indicating substantially reduced inflammation and tumour burden. Table 5The effect of nilotinib on serum tumor marker (ADA, AHH, and LDH) in control and treated groups.GroupsADA (pg/mL)AHH (ng/mL)LDH (U/L)Group I110.12 ± 4.52.85 ± 0.72118 ± 7.9Group II224.22 ± 5.25.90 ± 1.64158 ± 9.4Group III162.5 ± 2.333.76 ± 2.93841 ± 10.29Group IV149.68 ± 1.52.74 ± 0.332364 ± 5.2 Table 5The effect of nilotinib on serum tumor marker (ADA, AHH, and LDH) in control and treated groups.GroupsADA (pg/mL)AHH (ng/mL)LDH (U/L)Group I110.12 ± 4.52.85 ± 0.72118 ± 7.9Group II224.22 ± 5.25.90 ± 1.64158 ± 9.4Group III162.5 ± 2.333.76 ± 2.93841 ± 10.29Group IV149.68 ± 1.52.74 ± 0.332364 ± 5.2 The effect of nilotinib on serum tumor marker (ADA, AHH, and LDH) in control and treated groups. Serum tumour marker analysis (Table 5) exhibited marked elevation of ADA, AHH, and LDH in Group II compared to Group I controls, consistent with tumour induction. Nilotinib treatment produced dose-dependent reductions in all three markers, with Group IV (75 mg/kg) reaching values closest to control levels. Adenosine deaminase (ADA) regulates purine metabolism; aryl hydrocarbon hydroxylase (AHH) has been linked to tumour incidence and carcinogen activation; and lactate dehydrogenase (LDH) serves as a prognostic marker reflecting inflammation, tumour burden, and immunosuppression in NSCLC. Histopathological examination of lung sections (Fig. 16) confirmed that control mice displayed normal alveolar architecture and thin septa. B[a]P-treated mice (Group II) exhibited well-defined tumour masses with irregular papillae lined by hyperplastic epithelium, with peribronchiolar inflammatory cell infiltration consistent with B[a]P-induced carcinogenesis. Nilotinib at 50 mg/kg (Group III) showed thickened septa with residual inflammatory infiltrate, while treatment at 75 mg/kg (Group IV) produced marked thinning of the septa and significantly reduced inflammatory cell aggregation. Flow cytometry (FACS) analysis of immune cell populations in NSCLC T cell populations When comparing B[a]P-induced tumor-bearing mice to naïve controls, FACS analysis showed overexpression of CD3 + and CD8 + T cells, which is consistent with an adaptive immune response to tumor antigen. Significant decreases in CD3 + and CD8 + T cell populations were seen in nilotinib-treated mice (B[a]P + Nil) when compared to the B[a]P vehicle group (Fig. 17). Fig. 17Flow cytometric scatter plots and bar graphs showing CD3 + CD8+ and CD3 + CD4+T cell subsets in lung tissue from naïve controls, B[a]P + vehicle (B[a]P + VEH), and B[a]P + nilotinib (B[a]P + Nil) groups. n = 6 per group. Statistical significance was determined by one-way ANOVA with Dunnett’s multiple comparison test, ****p < 0.0001, *p < 0.05 vs. naïve control. Flow cytometric scatter plots and bar graphs showing CD3 + CD8+ and CD3 + CD4+T cell subsets in lung tissue from naïve controls, B[a]P + vehicle (B[a]P + VEH), and B[a]P + nilotinib (B[a]P + Nil) groups. n = 6 per group. Statistical significance was determined by one-way ANOVA with Dunnett’s multiple comparison test, ****p < 0.0001, *p < 0.05 vs. naïve control. Dendritic cells and monocytes Tumour-induced mice exhibited marked increases in CX3CR1 + CD11c+ dendritic cells and Ly6c + MHC II+ monocytes/macrophages compared with naïve controls, indicative of chronic tumour-associated immune activation. Nilotinib treatment significantly reduced both dendritic cell and monocyte populations (Figs. 18 and 19). CX3CR1 overexpression has been associated with an immunosuppressive tumour microenvironment through promotion of M2 macrophage polarisation; the reduction of CX3CR1 + CD11c+ cells following nilotinib treatment may therefore contribute to remodelling of the pro-tumorigenic immune landscape. Fig. 18Flow cytometric scatter plots and bar graphs showing CX3CR1 + CS11c+ dendritic cell populations in lung tissue from naïve controls, B[a]P + vehicle and B[a]P + nilotinib groups. n = 6 per group. Statistical significance was determined by one-way ANOVA with Dunnett’s multiple comparison test, ****p < 0.0001 vs. naïve control. Flow cytometric scatter plots and bar graphs showing CX3CR1 + CS11c+ dendritic cell populations in lung tissue from naïve controls, B[a]P + vehicle and B[a]P + nilotinib groups. n = 6 per group. Statistical significance was determined by one-way ANOVA with Dunnett’s multiple comparison test, ****p < 0.0001 vs. naïve control. Fig. 19Flow cytometric scatter plots and bar graphs showing Ly6c + MHC II+ monocyte/macrophage populations in lung tissue from naïve controls, B[a]P + vehicle, and B[a]P + nilotinib groups. n = 6 per group. Statistical significance was determined by one-way ANOVA with Dunnett’s multiple comparison test, ****p < 0.0001, ***p < 0.001 vs. naïve control. Flow cytometric scatter plots and bar graphs showing Ly6c + MHC II+ monocyte/macrophage populations in lung tissue from naïve controls, B[a]P + vehicle, and B[a]P + nilotinib groups. n = 6 per group. Statistical significance was determined by one-way ANOVA with Dunnett’s multiple comparison test, ****p < 0.0001, ***p < 0.001 vs. naïve control. M2 macrophage aggregation CX3CR1 overexpression in NSCLC was associated with a marked increase in M2 macrophage (CD206 + F4/80+) populations in tumour-bearing mice, confirming the immunosuppressive and pro-tumorigenic role of M2 polarisation. Nilotinib treatment significantly reduced the M2 macrophage fraction, particularly at the higher dose, suggesting that nilotinib may reshape the tumour immune microenvironment by attenuating M2-mediated immunosuppression. This finding is of potential translational relevance given the growing interest in combining targeted therapies with immune checkpoint blockade; preliminary evidence of nilotinib-mediated M2 suppression may indicate synergistic potential with anti-PD-1/PD-L1 agents, which warrants further investigation (Fig. 20). Fig. 20Flow cytometric scatter plots and bar graphs showing M2 macrophage (CD206 + F4/80+) populations in lung tissue from naïve controls, B[a]P + vehicle and B[a]P + nilotinib groups. n = 6 per group. Statistical significance was determined by one-way ANOVA with Dunnett’s multiple comparison test, ****p < 0.0001 vs. naïve control. Flow cytometric scatter plots and bar graphs showing M2 macrophage (CD206 + F4/80+) populations in lung tissue from naïve controls, B[a]P + vehicle and B[a]P + nilotinib groups. n = 6 per group. Statistical significance was determined by one-way ANOVA with Dunnett’s multiple comparison test, ****p < 0.0001 vs. naïve control. Discussion The current study demonstrates that structure-based drug repurposing can identify clinically approved multi-targeted agents for KRAS-mutant NSCLC, with nilotinib emerging as a lead candidate across an integrated computational to preclinical pipeline. Originally approved as a BCR-ABL/c-Kit/PDGFR inhibitor for Chronic myeloid leukemia (CML), nilotinib exhibited the strongest predicted binding affinity across all six MAPK pathway targets (EGFR, KRAS G12C, BRAF, MEK, ERK, and SHP2) with stable complex trajectories of MD simulation. Experimentally, nilotinib reduced NCI-H23 cell viability (IC₅₀ 31.51 µM), clonogenic survival (233 ± 7 colonies vs. 557 ± 12 untreated), and collective migration more effectively than cisplatin at 24 h, while inducing G2/M cell cycle accumulation and combined apoptosis of ~ 73%. Strong anti-angiogenic activity was confirmed in the ex ovo CAM model. In vivo, nilotinib normalised B[a]P-elevated serum tumour markers (ADA, AHH, LDH), reduced histological tumour burden, and modulated the tumour immune microenvironment by decreasing M2 macrophage, dendritic cell, and monocyte populations which together indicates multi-axis anti-tumour activity bridging proliferation, angiogenesis, and immune regulation. Several targeted drugs and kinase inhibitors have been investigated as potential repurposing candidates in NSCLC. Due to inadequate lung pharmacokinetics and poor KRAS G12C engagement, imatinib, a structurally related BCR-ABL/c-Kit inhibitor, has demonstrated limited single-agent NSCLC activity. Due to downstream resistance, approved EGFR inhibitors gefitinib and erlotinib are mostly ineffective in KRAS-driven disease but effective in EGFR-mutant NSCLC. Although it lacked KRAS-specific binding, sorafenib, a multi-kinase inhibitor that covers BRAF and VEGFR, showed some anti-angiogenic activity in NSCLC trials. Several computational repurposing studies have been reported for NSCLC, targeting proteins such as EGFR and Akt-1 using molecular docking and molecular dynamics simulations4,16,20. In addition, aberrant MAPK signalling remains a central driver of NSCLC progression and therapeutic resistance, particularly in tumours harbouring KRAS mutations7–12. While approved KRAS G12C inhibitors like sotorasib have shown significant clinical benefits by selectively targeting mutant KRAS22, resistance often develops through the reactivation of downstream signaling pathways. The wider target engagement profile expected for nilotinib suggests a possibly supportive treatment strategy that needs more investigation. Despite strong computational and experimental evidence supporting nilotinib’s anticancer potential in KRAS-mutant NSCLC, certain limitations persist. In contrast to selective KRAS G12C inhibitors, nilotinib was not made to specifically target mutant KRAS. The noted biological effects might come from its broader activity on kinases. This idea is backed by the significant drop in proliferation, migration, clonogenicity, and angiogenesis seen in several experimental models. Notably, the IC₅₀ value in NCI-H23 cells was higher than the potency of clinically approved KRAS G12C inhibitors. This suggests that direct comparisons with mutation-specific targeted therapies should be approached carefully. The covalent KRAS G12C interaction is a computational prediction. Nilotinib lacks a classical electrophilic warhead, and experimental confirmation via intact protein mass spectrometry, thermal shift assay and RAS-GTP pull-down has not yet been performed. The in vivo model lacks confirmed KRAS G12C driver mutations and therefore does not directly validate KRAS G12C-specific in vivo activity. Future work will prioritize biochemical confirmation of KRAS G1C engagement, validation in KRAS G12C-specific in vivo model (NCI-H23 xenograft in NSG mice), pharmacokinetic profiling at therapeutic doses and evaluation of nilotinib in combination with sotorasib or anti-PD-1/PD-L1 agents given the immunomodulatory activity identified here. Finally, while drug repurposing accelerates translational potential, further toxicity, pharmacokinetic, and combinatorial studies are essential before advancing to clinical evaluation. Conclusion This study establishes structure-based drug repurposing as a viable strategy for accelerating therapeutic discovery in KRAS-mutant NSCLC. Through an integrated computational pipeline and experimental validation, nilotinib emerged as a potent multi-targeted candidate with strong predicted affinity for the KRAS G12C binding site and key MAPK pathway regulators. In vitro assays confirmed its ability to inhibit proliferation, suppress clonogenic survival, reduce collective cell migration, induce apoptosis and cell cycle arrest accumulation in KRAS G12C mutant NCI-H23 cells. The ex ovo CAM assay further validated nilotinib’s anti-angiogenic activity and in vivo studies in the B[a]P-induced lung carcinogenesis model demonstrated significant reductions in tumour-associated markers and immune cell modulation in the tumour microenvironment. These findings underscore the translational potential of nilotinib beyond its approved indication in CML, supporting its further evaluation as a repurposed agent for KRAS-driven lung cancer. Critical next steps include biochemical confirmation for KRAS G12C engagement, validation in a KRAS G12C-specific in vivo model, pharmacokinetic characterization at therapeutic doses and evaluation of combination strategies with approved KRAS G12C inhibitors or immune checkpoint blockade agents. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1